[SOLVED] Sudden screeching noise out of Laptop 16 while running

Framework Laptop 16 Linux
1

Earlier while I was using my Framework Laptop 16 I got jumpscared by an extremely loud, continuous screeching noise. I thought the building’s fire alarm went off until I left my apartment and realized it was inside. I wrote down later that it sounded like a low siren or a turbine. The frequency would sometimes shift around, though not on a regular cycle. I know that the obvious lead is the fans; I want to make it clear that I’ve heard computer fans at full power before and this was much louder.

I scrambled for some 2 minutes trying to figure out the source as I have quite a few devices on my desk. I unplugged the main power strip (to no effect) before realizing the Framework Laptop 16 was the cause. I disconnected all peripherals which eventually stopped the noise as unplugging both displays while the laptop’s lid was closed (which is how I normally use it) puts it to sleep.

I know sleep mode is relevant because I later opened the lid back up and the noise restarted instantly. This means the noise was happening while the computer was on battery without anything plugged into it. At that point I was genuinely concerned about fire risks and pressed the power button, which shut it down normally.

Throughout the entire episode the system kept running, with both external displays and the internal one working when they should, and I made no note of anything looking wrong with the OS. Although I do not remember using any input device other than the power button. I also wrote down that the computer was not particularly hot and the case did not shock me when I inevitably touched it (at least nothing out of the ordinary).

Needless to say, this scared the shit out of me and I’m now genuinely afraid to turn it back on, which is not the kind of first experience I was counting on today :frowning:

Frustratingly, I don’t have any video or audio recording. As I’m sure you’ll understand, I was dead focused on stopping the literal alarm.

When this happened I had just connected the jack to my headphones (jack is always connected on the laptop side but I often disconnect it on the headphones side to use Bluetooth) and I was in pavucontrol about to switch the audio output there instead of speakers (my config is wonky so it doesn’t switch automatically). This is something I do all the time. In case this is relevant, I want to reiterate that unplugging the computer side of the jack did nothing about the noise while it was happening.

I’ve now taken the hard drive out (fsck is happy, which is at least something) for extra backups and restoring my system on another machine, from which I’m writing this post. I will try to get some logs out. I’m considering removing the battery and trying to use the Framework Laptop 16 again but I’d rather have some opinion first on how that noise could happen and if there is any danger.

Machine details: Ryzen 7 7840HS, BIOS version ??, no external graphics card
System details: Arch (last updated a week ago), latest kernel

Peripherals at the time of the incident (been using this setup for months):

  • Back left HDMI adapter to external display
  • Middle left DP adapter to external display
  • Front left USB-A hub with one device plugged in, turned off
  • Back right USB-C charger
  • Middle right Jack (connected)
  • Front right USB-A mouse
  • Charger plugged in, battery fully charged (I have used the laptop on battery recently and even allowed the battery to empty completely on accident a few days ago)
  • Lid closed
2

Listening to my substitution machine’s fan when it ramps up I can hear it has a faint siren-like quality too. So I would say there is a good chance that the Framework noise was the fans. Still it was like insanely loud (one of the patterns my brain matched while trying to figure out the noise was drilling a hole in the wall). Is the Framework beefy enough to do that? o_o

Attached is a photo of the midplate, in case anything is wrong. It looks good I think, in which case my speculation would be that either firmware or kernel suddenly went completely ham with the fans for no reason.

I’ve had old laptops with a hidden Fn shortcut for putting the fans at max speed. Since I use an external USB keyboard with no Fn key I assume that I couldn’t have hit such a shortcut if there is one on the Framework.

P1010575
P10105751920×1440 279 KB

3

It sounds to me like it should be safe to start it up again, just to see if the sound immediately starts again. If it’s a mechanical noise, it would almost have to be the fans, I don’t think the FW16 has any other mechanical devices in it. And if it’s from the fans, turning them on again shouldn’t be able to cause damage to any other parts.

If it was somehow coming through the speakers instead, it might have been caused by a cosmic-ray-induced bit-flip at just the wrong location in memory. I’ve never heard of that happening, but it might be technically possible.

2 Likes
4

Yeah, I’d try not to worry about the sound an indicator of any danger. Of course, devices plugged into a high current power supply and / or containing a li-ion battery are capable of fires, but I believe they’re really unlikely to warn you with a very loud siren-like noise.

The fan physically scraping something is most likely. Might only happen at a certain fan speed. It could take a particular vibration from a certain speed for an errant fan blade to just barely contact the fan case.

Oh, that’s something we tend to forget about. Cosmic-ray-induced bit-flips certainly happen, and being random, and anywhere, they could potentially cause a great variety of oddities.

2 Likes
5

Thank you both. I tried to boot again, first time with the midplate off, and the fans did spin normally, then a second time with the laptop fully assembled. For some reason some boot settings were lost (I didn’t remove the battery in the end). Anyway works fine, no huge noise the second time either. I have yet to try a heavy load.

I hope this isn’t the end of it because I don’t quite like the feeling that I can’t reproduce it, I don’t know when it’ll happen again, and I don’t really know what to do either if it does (apart from shut down and hope it goes away). Please keep any theories coming…

This indeed seems to match my observations best. I thought there could be some small debris but it started fine with my first boot above. Is there any way I could check for this case or help prevent it from happening?

It definitely sounded mechanical. That said, a bit flip would have been funny. I happened to be writing my Ph.D thesis on protections against fault injections when it happened. That would have made for a hilarious anecdote.

3 Likes
6

One very important rule of troubleshooting is to check the obvious.

  • There is no hard drive in a Framework 16, it only has room for solid state SSDs. No moving parts in an SSD that can make noises.
  • I don’t think the battery could make a noise either. Not without some obvious deformations. So no reason to remove that either.
  • Obvious sources of noise could be the fans or speakers. So first thing I would do is remove and inspect the speakers. There was someone else in this forum who had basically a speaker blowout and their speaker was deformed in shape.
  • It is also a possibility some software was outputting the noise. Either a glitch/bug or some type of malware.
1 Like
7

Thanks. The speakers seem alright and still work.

But now with the extra input I think I know what might have been the root cause. Which would be just the right mix of buggy behaviors and shameful incompetence on my part. :weary_face:

Here’s my theory. When I switched the audio in pavucontrol, I remember going to select the jack but it wasn’t where I expected it to be, so I clicked back on the speakers. Now my audio stack uses Pipewire with the Pulseaudio emulation. Due to either bugs or just extreme friction with the tools, whenever the audio sink switches Pipewire reconfigures a bunch of stuff, usually duplicating controls, rerouting inputs to outputs and whatnot. It usually doesn’t make any difference, usually.

I’d been on a call recently where I set up a combined sink with my microphone and an application source so I could stream my app sound to the call. The setup to do that is somewhat convoluted and involves a PulseAudio loopback module. It’s too late to check but I don’t remember unloading all the modules after the call.

You might see where I’m going. I think that when I switched the audio sink Pipewire decided to connect the forgotten loopback module (which was set up to capture the microphone) to the new output (even though I had it connected only to the virtual mic). This is stupid, but it had happened before. And then of course echo mic to speakers and queue the catastrophic feedback loop. :man_facepalming:

With a quickly rising volume the mic would have been on at all times, bypassing any background noise suppression, thus picking up the sound of the fans (especially with the lid closed) which the speakers would amplify. Hence my perception of super loud fans.

1 Like
8

No good way to test comes to mind. I forget if there is an easy way to control fan speed in Linux, exact speed. If there is, very slowly ramping speed from zero to full would be worth a shot.

Since the fans are more accessible in the FWL16 you could clean the blades over a sheet of plastic to catch what comes out, looking for a piece of stiff or hard debris. Plus, you’ll come away with slightly better cooling for your efforts.

1 Like
9

That could sure provide quite a screeching noise. Hurts my head just thinking of it. :face_with_spiral_eyes:

1 Like
10

After thinking it over I’m satisfied with the Pulseaudio loop explanation, as it explains the sound, the context when it happened, and the fact nothing was broken. Plus I feel more likely to mess up my audio setup than for Framework to mess up with the laptop.

I’d edit the top post to mark as resolved and indicate this conclusion, but I can’t. If any mods around can do it, that’d be much appreciated. Thanks everyone for the help diagnosing.

11

I think you can only mark a particular post as a solution if the thread is in the Community Support subforum? I’m not certain. I don’t make a lot of threads :upside_down_face:

12

Well, you do seem to have sorted the noise problem out, but another thing to consider when dealing with such a noise is dry bearings in a fan. This would normally happen on a very old machine which hasn’t had any maintenance or cleaning and the oil in the bearings dries out, and can produce a very loud screeching noise. But that tends to happen at turn on and for a limited time.

As to bit flips, back in the early 1980s I was involved in tech support for a minicomputer system that used the earliest 16kx1 dynamic RAM chips which required three supplies (+12v, +5v, -5v) to operate. These were mounted on a plug in daughter board so were easy to change or add modules to increase the memory. Periodically I would get one back from the field with ‘permanent’ parity errors, and on test it would consistently give an error on one bit on the module. If then left on the shelf unpowered for about a week the parity error would go away. The theory was that an alpha particle (or it might have been a beta one, i don’t remember which) would hit one of the chips in just the right place to lodge in the floating storage capacitor that stored the state of that bit. One particle caused a charge equivalent to about 16000 electrons, which was about what the cell required to set the state of that bit. If the board was left unpowered the charge would eventually leak away and that bit would operate normally once more, but if powered up each day the charge would not leak away. The source of the particle causing the problem was unknown, could have been cosmic, but the ceramics used in the chip packages at the time were also known to have impurities which could emit atomic particles.